Natural Language User Profiles for Transparent and Scrutable Recommendations

Current state-of-the-art recommender systems predominantly rely on either implicit or explicit feedback from users to suggest new items. While effective in recommending novel options, these conventional systems often use uninterpretable embeddings. This lack of transparency not only limits user understanding of why certain items are suggested but also reduces the user's ability to easily scrutinize and edit their preferences. For example, if a user has a change in interests, they would need to make significant changes to their interaction history to adjust the model's recommendations. To address these limitations, we introduce a novel method that utilizes user reviews to craft personalized, natural language profiles describing users' preferences. Through these descriptive profiles, our system provides transparent recommendations in natural language. Our evaluations show that this novel approach maintains a performance level on par with established recommender systems, but with the added benefits of transparency and user control. By enabling users to scrutinize why certain items are recommended, they can more easily verify, adjust, and have greater autonomy over their recommendations.